A New Perspective on Machine Learning: How to do Perfect Supervised Learning
This provides a theoretical foundation to explain the success of large-scale supervised learning with complex models, potentially impacting all of ML/AI.
The paper tackles the problem of achieving perfect supervised learning by introducing bandlimiting constraints, showing that practical tasks become asymptotically solvable with enough training samples and sufficiently complex, bandlimited models, and deriving new error bounds that quantify learning difficulty.
In this work, we introduce the concept of bandlimiting into the theory of machine learning because all physical processes are bandlimited by nature, including real-world machine learning tasks. After the bandlimiting constraint is taken into account, our theoretical analysis has shown that all practical machine learning tasks are asymptotically solvable in a perfect sense. Furthermore, the key towards this solvability almost solely relies on two factors: i) a sufficiently large amount of training samples beyond a threshold determined by a difficulty measurement of the underlying task; ii) a sufficiently complex and bandlimited model. Moreover, for some special cases, we have derived new error bounds for perfect learning, which can quantify the difficulty of learning. These generalization bounds are not only asymptotically convergent but also irrelevant to model complexity. Our new results on generalization have provided a new perspective to explain the recent successes of large-scale supervised learning using complex models like neural networks.